We have assessed the effect of different regimes of antibiotic prophylaxis on the survival of total hip implants, comparing antibiotics administered both systemically and in the bone cement, systemically only, in the bone cement only and with no antibiotics given. We studied 10 905 primary cemented total hip replacements, performed for osteoarthritis of the hip and reported to the Norwegian arthroplasty register between 1987 and 1995. Cox-estimated failure-rate ratios (FRR) are presented with adjustment for gender, age, the brand of cement, the prosthesis, the type of operating theatre and the operating time. For revisions performed for infection (39 operations), the lowest rate of revision was found among patients receiving antibiotic-containing cement plus systemic antibiotics (n = 5804). The revision rate for the 4586 patients receiving systemic antibiotics only was 4.3 times greater (95% CI 1.7 to 11.0, p = 0.001); in 239 with antibiotics in the bone cement only it was 6.3 times greater (CI 1.6 to 25.0, p = 0.003); and in the 276 who did not receive antibiotics it was by 11.5 times greater (CI 2.1 to 63.0, p = 0.002). Adjustment for the total amount of systemic antibiotic administered did not change the results. We also observed an increased revision rate for aseptic loosening (109 operations) comparing the systemic-only (FRR = 1.8, CI 1.1 to 2.9, p= 0.01) and the cement-only regimes (FRR = 2.6, CI 1.2 to 5.9, p = 0.02) with the combined dosage. Our findings show that systemic antibiotics combined with antibiotic-containing bone cement led to fewer revisions than the other methods.